A video processing method is provided to include performing a conversion between a coded representation of a video including one or more video regions and the video, wherein the coded representation includes side information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the side information provides parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the side information applicable to the some of the one or more video regions is coded without directly inheriting from other video regions.
|
1. A method of processing video data, comprising:
determining whether an in-loop reshaping (ILR) is enabled for a conversion between a bitstream of a video comprising one or more video regions and the video;
reconstructing a video unit of a video region by applying information associated with the ILR in a case that the determining determines that the ILR is enabled, the reconstructing of the video unit being based on a bitstream of the video unit in a first domain and a second domain; and
performing the conversion by utilizing fixed clipping parameters unrelated to the information associated with the ILR in a case that the determining determines that the ILR is disabled and utilizing ILR clipping parameters derived according to the information associated with the ILR in a case that the determining determines that the ILR is enabled;
wherein in the case that the determining determines that the ILR is enabled, the ILR clipping parameters represented in the first domain and/or the second domain are used for the conversion.
17. A non-transitory computer-readable storage medium storing instructions that cause a processor to:
determine whether an in-loop reshaping (ILR) is enabled for a conversion between a bitstream of a video comprising one or more video regions and the video;
reconstruct a video unit of a video region by applying information associated with the ILR in a case that the determining determines that the ILR is enabled, the reconstructing of the video unit being based on a bitstream of the video unit in a first domain and a second domain; and
perform the conversion by utilizing fixed clipping parameters unrelated to the information associated with the ILR in a case that the determining determines that the ILR is disabled and utilizing ILR clipping parameters derived according to the information associated with the ILR in a case that the determining determines that the ILR is enabled;
wherein in the case that the determining determines that the ILR is enabled, the ILR clipping parameters represented in the first domain and/or the second domain are used for the conversion.
16. An apparatus for processing video data comprising a processor and a non-transitory memory with instructions thereon, wherein the instructions upon execution by the processor, cause the processor to:
determine whether an in-loop reshaping (ILR) is enabled for a conversion between a bitstream of a video comprising one or more video regions and the video;
wherein a clipping process is used for the conversion based on ILR clipping parameters reconstruct a video unit of a video region by applying information associated with the ILR in a case that the determining determines that the ILR is enabled, the reconstructing of the video unit being based on a bitstream of the video unit in a first domain and a second domain; and
perform the conversion by utilizing fixed clipping parameters unrelated to the information associated with the ILR in a case that the determining determines that the ILR is disabled and utilizing ILR clipping parameters derived according to the information associated with the ILR in a case that the determining determines that the ILR is enabled;
wherein in the case that the determining determines that the ILR is enabled, the ILR clipping parameters represented in the first domain and/or the second domain are used for the conversion.
2. The method of
3. The method of
the ILR clipping parameters are used during a prediction block generation process, the prediction block generation process including at least one of intra and/or intra block copy generation or inter prediction block generation;
or, the ILR clipping parameters are used during the conversion performed in a palette mode in which at least a palette of representative sample values is used for the video unit;
or, the ILR clipping parameters are used during a motion compensation process;
or, the ILR clipping parameters are used during a final reconstruction block derivation process;
or, wherein the ILR clipping parameters are used during a filtering process.
4. The method of
5. The method of
the ILR clipping parameters are defined in the second domain and used for the conversion before an ILR filtering operation,
or, the ILR clipping parameters are defined in the first domain and used during an in-loop filtering operation applied to the video unit.
6. The method of
7. The method of
8. The method of
determining a prediction block for the conversion using samples from the video region, and
determining unavailable samples of the samples from the video region based on a look up table or a piece-wise linear (PWL) model used in the ILR.
9. The method of
detecting a certain condition associated with the video unit,
wherein the certain condition is whether the ILR is enabled or disabled for a video region including the video unit,
or, wherein the certain condition is related to any one or more of: a luma color component, a main color component, a color format, or a chroma component.
10. The method of
11. The method of
12. The method of
13. The method of
|
This application is a continuation of International Patent Application No. PCT/CN2020/077725, filed on Mar. 4, 2020, which claims the priority to and benefits of International Patent Application No. PCT/CN2019/076852, filed on Mar. 4, 2019, International Patent Application No. PCT/CN2019/077643, filed on Mar. 11, 2019, and International Patent Application No. PCT/CN2019/078410, filed on Mar. 17, 2019. All of the aforementioned applications are incorporated by reference as part of the disclosure of this application.
This patent document relates to video processing techniques, devices and systems.
Currently, efforts are underway to improve the performance of current video codec technologies to provide better compression ratios or provide video coding and decoding schemes that allow for lower complexity or parallelized implementations. Industry experts have recently proposed several new video processing tools and tests are currently underway for determining their effectivity.
Devices, systems and methods related to digital video processing, for example, block-based in-loop reshaping with other tools in video processing. It may be applied to the existing video coding standard like HEVC, or the standard (Versatile Video Coding) to be finalized. It may be also applicable to future video coding standards or video codec.
In one representative aspect, the disclosed technology may be used to provide a method for video processing. This method includes performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes side information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the side information provides parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the side information applicable to the some of the one or more video regions is coded without directly inheriting from other video regions.
In another representative aspect, the disclosed technology may be used to provide another method for video processing. This method includes performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes side information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the side information provides parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the coded representation includes the side information at a video region level that includes a current video block.
In another representative aspect, the disclosed technology may be used to provide another method for video processing. This method includes determining a prediction block of a current video block, for a conversion between the current video block of a video region of a video and a coded representation of the video, using samples from the video region, wherein unavailable samples of the samples from the video region are determined based on a derivation rule that specifies to use an in-loop reshaping (ILR) information associated with the current video block; and performing the conversion based on the prediction block, wherein the in-loop reshaping information includes information for a reconstruction of a video unit of the video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit.
In another representative aspect, the disclosed technology may be used to provide another method for video processing. This method includes performing a conversion between a current video block of a video and a coded representation of the video using an in loop reshaping process, wherein the in loop reshaping process includes: obtaining a prediction block from motion compensation prediction or an intra prediction; applying a forward reshaping to the prediction block to generate a forward reshaped prediction block in a first domain; reconstructing, based on the forward reshaped prediction block, the coded representation of the video to generate a first reconstruction block in a second domain; applying an inverse reshaping to the first reconstruction block to obtain an inverse reconstruction block; applying a loop filter to the inverse reconstruction block to obtain a final reconstruction block, and wherein a single clipping process is used during the in loop reshaping process.
In another representative aspect, the disclosed technology may be used to provide another method for video processing. This method includes performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes in-loop reshaping (ILR) information applicable for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein a clipping process is used for the conversion based on clipping parameters that are in the first domain and/or in the second domain according to the ILR information.
In another representative aspect, the disclosed technology may be used to provide another method for video processing. This method includes performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the information provides ILR parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the ILR parameters include clipping parameters used according to a rule for a clipping process during the rule.
In another representative aspect, the disclosed technology may be used to provide another method for video processing. This method includes determining whether an in loop reshaping (ILR) is enabled for a conversion between a coded representation of a video comprising one or more video regions and the video; and performing the conversion based on the determining, wherein a clipping process is used for the conversion based on ILR clipping parameters derived according to information associated with the ILR and/or fixed clipping parameters unrelated to the information associated with the ILR, and wherein the information associated with the ILR is applicable for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit.
In another representative aspect, the disclosed technology may be used to provide another method for video processing. This method includes performing a conversion between a current video block of a video that is a chroma block and a coded representation of the video, wherein the coded representation includes side information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the side information provides parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the conversion further includes applying a reshaping process to one or more chroma components of the current video block based on the side information.
In another representative aspect, the disclosed technology may be used to provide another method for video processing. This method includes performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes side information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the side information provides parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the side information for a current video region is determined according to a rule that disallows inheritance or prediction of the side information based on a temporal layer index of the current video region.
In another representative aspect, the disclosed technology may be used to provide another method for video processing. This method includes performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes side information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the side information provides parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the side information for a current video region is determined according to a rule that disallows inheritance or prediction of the side information from a first type of another picture.
In another representative aspect, the disclosed technology may be used to provide another method for video processing. This method includes performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes a first side information at a first level, wherein a second side information at a second level is derived from the first side information such that the second side information provides parameters for a video unit coded with in-loop reshaping (ILR) in which a reconstruction of the video unit of a video region is based on a representation of a video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit.
One or more of the above-disclosed methods can be an encoder-side implementation or a decoder-side implementation.
Further, in a representative aspect, an apparatus in a video system comprising a processor and a non-transitory memory with instructions thereon is disclosed. The instructions upon execution by the processor, cause the processor to implement any one or more of the disclosed methods.
Also, a computer program product stored on a non-transitory computer readable media, the computer program product including program code for carrying out any one or more of the disclosed methods is disclosed.
The above and other aspects and features of the disclosed technology are described in greater detail in the drawings, the description and the claims.
Video coding standards have evolved primarily through the development of the well-known ITU-T and ISO/IEC standards. The ITU-T produced H.261 and H.263, ISO/IEC produced MPEG-1 and MPEG-4 Visual, and the two organizations jointly produced the H.262/MPEG-2 Video and H.264/MPEG-4 Advanced Video Coding (AVC) and H.265/HEVC standards. Since H.262, the video coding standards are based on the hybrid video coding structure wherein temporal prediction plus transform coding are utilized. To explore the future video coding technologies beyond HEVC, Joint Video Exploration Team (JVET) was founded by VCEG and MPEG jointly in 2015. Since then, many new methods have been adopted by JVET and put into the reference software named Joint Exploration Model (JEM). In April 2018, the Joint Video Expert Team (JVET) between VCEG (Q6/16) and ISO/IEC JTC1 SC29/WG11 (MPEG) was created to work on the VVC standard targeting at 50% bitrate reduction compared to HEVC. The latest version of VVC draft, i.e., Versatile Video Coding (Draft 2) could be found at http://phenix.it-sudparis.eu/jvet/doc_end_user/documents/11_Ljubljana/wg11/JVET-K1001-v7.zip The latest reference software of VVC, named VTM, could be found at: https://vcgit.hhi.fraunhofer.de/jvet/VVCSoftware_VTM/tags/VTM-2.1
2.1. Inter Prediction in HEVC/H.265
Each inter-predicted PU has motion parameters for one or two reference picture lists. Motion parameters include a motion vector and a reference picture index. Usage of one of the two reference picture lists may also be signaled using inter_pred_idc. Motion vectors may be explicitly coded as deltas relative to predictors.
When a CU is coded with skip mode, one PU is associated with the CU, and there are no significant residual coefficients, no coded motion vector delta or reference picture index. A merge mode is specified whereby the motion parameters for the current PU are obtained from neighboring PUs, including spatial and temporal candidates. The merge mode can be applied to any inter-predicted PU, not only for skip mode. The alternative to merge mode is the explicit transmission of motion parameters, where motion vector (to be more precise, motion vector differences (MVD) compared to a motion vector predictor), corresponding reference picture index for each reference picture list and reference picture list usage are signaled explicitly per each PU. Such a mode is named Advanced motion vector prediction (AMVP) in this disclosure.
When signaling indicates that one of the two reference picture lists is to be used, the PU is produced from one block of samples. This is referred to as “uni-prediction.” Uni-prediction is available both for P-slices and B-slices.
When signaling indicates that both of the reference picture lists are to be used, the PU is produced from two blocks of samples. This is referred to as ‘bi-prediction’. Bi-prediction is available for B-slices only.
The following text provides the details on the inter prediction modes specified in HEVC. The description will start with the merge mode.
2.1.1. Reference Picture List
In HEVC, the term inter prediction is used to denote prediction derived from data elements (e.g., sample values or motion vectors) of reference pictures other than the current decoded picture. Like in H.264/AVC, a picture can be predicted from multiple reference pictures. The reference pictures that are used for inter prediction are organized in one or more reference picture lists. The reference index identifies which of the reference pictures in the list should be used for creating the prediction signal.
A single reference picture list, List 0, is used for a P slice and two reference picture lists, List 0 and List 1 are used for B slices. It should be noted reference pictures included in List 0/1 could be from past and future pictures in terms of capturing/display order.
2.1.2. Merge Mode
2.1.2.1. Derivation of Candidates for Merge Mode
When a PU is predicted using merge mode, an index pointing to an entry in the merge candidates list is parsed from the bitstream and used to retrieve the motion information. The construction of this list is specified in the HEVC standard and can be summarized according to the following sequence of steps:
In the following, the operations associated with the aforementioned steps are detailed.
2.1.2.2. Spatial Candidates Derivation
In the derivation of spatial merge candidates, a maximum of four merge candidates are selected among candidates located in the positions depicted in
2.1.2.3. Temporal Candidates Derivation
In this step, only one candidate is added to the list. Particularly, in the derivation of this temporal merge candidate, a scaled motion vector is derived based on co-located PU belonging to the picture which has the smallest POC difference with current picture within the given reference picture list. The reference picture list to be used for derivation of the co-located PU is explicitly signaled in the slice header. The scaled motion vector for temporal merge candidate is obtained as illustrated by the dotted line in
In the co-located PU (Y) belonging to the reference frame, the position for the temporal candidate is selected between candidates C0 and C1, as depicted in
2.1.2.4. Additional Candidates Insertion
Besides spatial and temporal merge candidates, there are two additional types of merge candidates: combined bi-predictive merge candidate and zero merge candidate. Combined bi-predictive merge candidates are generated by utilizing spatial and temporal merge candidates. Combined bi-predictive merge candidate is used for B-Slice only. The combined bi-predictive candidates are generated by combining the first reference picture list motion parameters of an initial candidate with the second reference picture list motion parameters of another. If these two tuples provide different motion hypotheses, they will form a new bi-predictive candidate. As an example,
Zero motion candidates are inserted to fill the remaining entries in the merge candidates list and therefore hit the MaxNumMergeCand capacity. These candidates have zero spatial displacement and a reference picture index which starts from zero and increases every time a new zero motion candidate is added to the list. Finally, no redundancy check is performed on these candidates.
2.1.3. AMVP
AMVP exploits spatio-temporal correlation of motion vector with neighboring PUs, which is used for explicit transmission of motion parameters. For each reference picture list, a motion vector candidate list is constructed by firstly checking availability of left, above temporally neighboring PU positions, removing redundant candidates and adding zero vector to make the candidate list to be constant length. Then, the encoder can select the best predictor from the candidate list and transmit the corresponding index indicating the chosen candidate. Similarly, with merge index signaling, the index of the best motion vector candidate is encoded using truncated unary. The maximum value to be encoded in this case is 2 (see
2.1.3.1. Derivation of AMVP Candidates
In motion vector prediction, two types of motion vector candidates are considered: spatial motion vector candidate and temporal motion vector candidate. For spatial motion vector candidate derivation, two motion vector candidates are eventually derived based on motion vectors of each PU located in five different positions as depicted in
For temporal motion vector candidate derivation, one motion vector candidate is selected from two candidates, which are derived based on two different co-located positions. After the first list of spatio-temporal candidates is made, duplicated motion vector candidates in the list are removed. If the number of potential candidates is larger than two, motion vector candidates whose reference picture index within the associated reference picture list is larger than 1 are removed from the list. If the number of spatio-temporal motion vector candidates is smaller than two, additional zero motion vector candidates is added to the list.
2.1.3.2. Spatial Motion Vector Candidates
In the derivation of spatial motion vector candidates, a maximum of two candidates are considered among five potential candidates, which are derived from PUs located in positions as depicted in
The no-spatial-scaling cases are checked first followed by the spatial scaling. Spatial scaling is considered when the POC is different between the reference picture of the neighboring PU and that of the current PU regardless of reference picture list. If all PUs of left candidates are not available or are intra coded, scaling for the above motion vector is allowed to help parallel derivation of left and above MV candidates. Otherwise, spatial scaling is not allowed for the above motion vector.
In a spatial scaling process, the motion vector of the neighboring PU is scaled in a similar manner as for temporal scaling, as depicted as
2.1.3.3. Temporal Motion Vector Candidates
Apart for the reference picture index derivation, all processes for the derivation of temporal merge candidates are the same as for the derivation of spatial motion vector candidates (see
2.2. Sub-CU Based Motion Vector Prediction Methods in JEM
In the JEM with QTBT, each CU can have at most one set of motion parameters for each prediction direction. Two sub-CU level motion vector prediction methods are considered in the encoder by splitting a large CU into sub-CUs and deriving motion information for all the sub-CUs of the large CU. Alternative temporal motion vector prediction (ATMVP) method allows each CU to fetch multiple sets of motion information from multiple blocks smaller than the current CU in the collocated reference picture. In the spatio-temporal motion vector prediction (STMVP) method, motion vectors of the sub-CUs are derived recursively by using the temporal motion vector predictor and spatial neighboring motion vector.
2.2.1. Alternative Temporal Motion Vector Prediction
ATMVP predicts the motion vectors of the sub-CUs within a CU in two steps. The first step is to identify the corresponding block in a reference picture with a so-called temporal vector. The reference picture is called the motion source picture. The second step is to split the current CU into sub-CUs and obtain the motion vectors as well as the reference indices of each sub-CU from the block corresponding to each sub-CU.
In the first step, a reference picture and the corresponding block is determined by the motion information of the spatial neighboring blocks of the current CU. To avoid the repetitive scanning process of neighboring blocks, the first merge candidate in the merge candidate list of the current CU is used. The first available motion vector as well as its associated reference index are set to be the temporal vector and the index to the motion source picture. This way, in ATMVP, the corresponding block may be more accurately identified, compared with TMVP, wherein the corresponding block (sometimes called collocated block) is always in a bottom-right or center position relative to the current CU.
In the second step, a corresponding block of the sub-CU is identified by the temporal vector in the motion source picture, by adding to the coordinate of the current CU the temporal vector. For each sub-CU, the motion information of its corresponding block (the smallest motion grid that covers the center sample) is used to derive the motion information for the sub-CU. After the motion information of a corresponding N×N block is identified, it is converted to the motion vectors and reference indices of the current sub-CU, in the same way as TMVP of HEVC, wherein motion scaling and other procedures apply. For example, the decoder checks whether the low-delay condition (i.e. the POCs of all reference pictures of the current picture are smaller than the POC of the current picture) is fulfilled and possibly uses motion vector MVx (the motion vector corresponding to reference picture list X) to predict motion vector MVy (with X being equal to 0 or 1 and Y being equal to 1-X) for each sub-CU.
2.2.2. Spatio-Temporal Motion Vector Prediction (STMVP)
In this method, the motion vectors of the sub-CUs are derived recursively, following raster scan order.
The motion derivation for sub-CU A starts by identifying its two spatial neighbors. The first neighbor is the N×N block above sub-CU A (block c). If this block c is not available or is intra coded the other N×N blocks above sub-CU A are checked (from left to right, starting at block c). The second neighbor is a block to the left of the sub-CU A (block b). If block b is not available or is intra coded other blocks to the left of sub-CU A are checked (from top to bottom, staring at block b). The motion information obtained from the neighboring blocks for each list is scaled to the first reference frame for a given list. Next, temporal motion vector predictor (TMVP) of sub-block A is derived by following the same procedure of TMVP derivation as specified in HEVC. The motion information of the collocated block at location D is fetched and scaled accordingly. Finally, after retrieving and scaling the motion information, all available motion vectors (up to 3) are averaged separately for each reference list. The averaged motion vector is assigned as the motion vector of the current sub-CU.
2.2.3. Sub-CU Motion Prediction Mode Signaling
The sub-CU modes are enabled as additional merge candidates and there is no additional syntax element necessary to signal the modes. Two additional merge candidates are added to merge candidates list of each CU to represent the ATMVP mode and STMVP mode. Up to seven merge candidates are used, if the sequence parameter set indicates that ATMVP and STMVP are enabled. The encoding logic of the additional merge candidates is the same as for the merge candidates in the HM, which means, for each CU in P or B slice, two more RD checks is needed for the two additional merge candidates.
In the JEM, all bins of merge index is context-coded by CABAC. While in HEVC, only the first bin is context coded and the remaining bins are context by-pass coded.
2.3. Local Illumination Compensation in JEM
Local Illumination Compensation (LIC) is based on a linear model for illumination changes, using a scaling factor a and an offset b. And it is enabled or disabled adaptively for each inter-mode coded coding unit (CU).
When LIC applies for a CU, a least square error method is employed to derive the parameters a and b by using the neighboring samples of the current CU and their corresponding reference samples. More specifically, as illustrated in
2.2.4. Derivation of Prediction Blocks
The IC parameters are derived and applied for each prediction direction separately. For each prediction direction, a first prediction block is generated with the decoded motion information, then a temporary prediction block is obtained via applying the LIC model. Afterwards, the two temporary prediction blocks are utilized to derive the final prediction block.
When a CU is coded with merge mode, the LIC flag is copied from neighboring blocks, in a way similar to motion information copy in merge mode; otherwise, an LIC flag is signaled for the CU to indicate whether LIC applies or not.
When LIC is enabled for a picture, additional CU level RD check is needed to determine whether LIC is applied or not for a CU. When LIC is enabled for a CU, mean-removed sum of absolute difference (MR-SAD) and mean-removed sum of absolute Hadamard-transformed difference (MR-SATD) are used, instead of SAD and SATD, for integer pel motion search and fractional pel motion search, respectively.
To reduce the encoding complexity, the following encoding scheme is applied in the JEM.
There are several new coding tools for inter prediction improvement, such as Adaptive motion vector difference resolution (AMVR) for signaling MVD, affine prediction mode, Triangular prediction mode (TPM), ATMVP, Generalized Bi-Prediction (GBI), Bi-directional Optical flow (BIO).
2.3.1. Coding Block Structure in VVC
In VVC, a QuadTree/BinaryTree/MulitpleTree (QT/BT/TT) structure is adopted to divide a picture into square or rectangle blocks.
Besides QT/BT/TT, separate tree (a/k/a Dual coding tree) is also adopted in VVC for I-frames. With separate tree, the coding block structure are signaled separately for the luma and chroma components.
2.3.2 Adaptive Motion Vector Difference Resolution
In HEVC, motion vector differences (MVDs) (between the motion vector and predicted motion vector of a PU) are signaled in units of quarter luma samples when use_integer_mv_flag is equal to 0 in the slice header. In the VVC, a locally adaptive motion vector resolution (AMVR) is introduced. In the VVC, MVD can be coded in units of quarter luma samples, integer luma samples or four luma samples (i.e., ¼-pel, 1-pel, 4-pel). The MVD resolution is controlled at the coding unit (CU) level, and MVD resolution flags are conditionally signaled for each CU that has at least one non-zero MVD components.
For a CU that has at least one non-zero MVD components, a first flag is signaled to indicate whether quarter luma sample MV precision is used in the CU. When the first flag (equal to 1) indicates that quarter luma sample MV precision is not used, another flag is signaled to indicate whether integer luma sample MV precision or four luma sample MV precision is used.
When the first MVD resolution flag of a CU is zero, or not coded for a CU (meaning all MVDs in the CU are zero), the quarter luma sample MV resolution is used for the CU. When a CU uses integer-luma sample MV precision or four-luma-sample MV precision, the MVPs in the AMVP candidate list for the CU are rounded to the corresponding precision.
2.3.3 Affine Motion Compensation Prediction
In HEVC, only translation motion model is applied for motion compensation prediction (MCP). While in the real world, there are many kinds of motion, e.g. zoom in/out, rotation, perspective motions and the other irregular motions. In VVC, a simplified affine transform motion compensation prediction is applied with 4-parameter affine model and 6-parameter affine model. As shown
The motion vector field (MVF) of a block is described by the following equations with the 4-parameter affine model (wherein the 4-parameter are defined as the variables a, b, e and f) in equation (1) and 6-parameter affine model (wherein the 4-parameter are defined as the variables a, b, c, d, e and f) in equation (2) respectively:
where (mvh0, mvh0) is motion vector of the top-left corner control point, and (mvh1, mvh1) is motion vector of the top-right corner control point and (mvh2, mvh2) is motion vector of the bottom-left corner control point, all of the three motion vectors are called control point motion vectors (CPMV), (x, y) represents the coordinate of a representative point relative to the top-left sample within current block and (mvh(x,y), mvv(x,y)) is the motion vector derived for a sample located at (x, y). The CP motion vectors may be signaled (like in the affine AMVP mode) or derived on-the-fly (like in the affine merge mode). w and h are the width and height of the current block. In practice, the division is implemented by right-shift with a rounding operation. In VTM, the representative point is defined to be the center position of a sub-block, e.g., when the coordinate of the left-top corner of a sub-block relative to the top-left sample within current block is (xs, ys), the coordinate of the representative point is defined to be (xs+2, ys+2). For each sub-block (i.e., 4×4 in VTM), the representative point is utilized to derive the motion vector for the whole sub-block.
In order to further simplify the motion compensation prediction, sub-block based affine transform prediction is applied. To derive motion vector of each M×N (both M and N are set to 4 in current VVC) sub-block, the motion vector of the center sample of each sub-block, as shown in
After MCP, the high accuracy motion vector of each sub-block is rounded and saved as the same accuracy as the normal motion vector.
2.3.3.1. Signaling of Affine Prediction
Similar to the translational motion model, there are also two modes for signaling the side information due affine prediction. They are AFFINE_INTER and AFFINE_MERGE modes.
2.3.3.2. AF_INTER Mode
For CUs with both width and height larger than 8, AF_INTER mode can be applied. An affine flag in CU level is signaled in the bitstream to indicate whether AF_INTER mode is used.
In this mode, for each reference picture list (List 0 or List 1), an affine AMVP candidate list is constructed with three types of affine motion predictors in the following order, wherein each candidate includes the estimated CPMVs of the current block. The differences of the best CPMVs found at the encoder side (such as mv0 mv1 mv2 in
1) Inherited Affine Motion Predictors
The checking order is similar to that of spatial MVPs in HEVC AMVP list construction. First, a left inherited affine motion predictor is derived from the first block in {A1, A0} that is affine coded and has the same reference picture as in current block. Second, an above inherited affine motion predictor is derived from the first block in {B1, B0, B2} that is affine coded and has the same reference picture as in current block. The five blocks A1, A0, B1, B0, B2 are depicted in
Once a neighboring block is found to be coded with affine mode, the CPMVs of the coding unit covering the neighboring block are used to derive predictors of CPMVs of current block. For example, if A1 is coded with non-affine mode and A0 is coded with 4-parameter affine mode, the left inherited affine MV predictor will be derived from A0. In this case, the CPMVs of a CU covering A0, as denoted by MV0N for the top-left CPMV and MV1N for the top-right CPMV in
2) Constructed Affine Motion Predictors
A constructed affine motion predictor consists of control-point motion vectors (CPMVs) that are derived from neighboring inter coded blocks, as shown in
No pruning process is applied when inserting a constructed affine motion predictor into the candidate list.
3) Normal AMVP Motion Predictors
The following applies until the number of affine motion predictors reaches the maximum.
In AF_INTER mode, when 4/6-parameter affine mode is used, 2/3 control points can be used, and therefore 2/3 MVD needs to be coded for these control points, as shown in
mv0=
mv1=
mv2=
wherein
2.3.3.3. AF_MERGE Mode
When a CU is applied in AF_MERGE mode, it gets the first block coded with affine mode from the valid neighbour reconstructed blocks. And the selection order for the candidate block is from left, above, above right, left bottom to above left as shown in
After the CPMV of the current CU mv0C, mv1C and mv2C are derived, according to the simplified affine motion model Equation (1) and (2), the MVF of the current CU is generated. In order to identify whether the current CU is coded with AF_MERGE mode, an affine flag is signaled in the bitstream when there is at least one neighbour block is coded in affine mode.
In JVET-L0142 and JVET-L0632, an affine merge candidate list is constructed with following steps:
1) Insert Inherited Affine Candidates
Inherited affine candidate means that the candidate is derived from the affine motion model of its valid neighbor affine coded block. The maximum two inherited affine candidates are derived from affine motion model of the neighboring blocks and inserted into the candidate list. For the left predictor, the scan order is {A0, A1}; for the above predictor, the scan order is {B0, B1, B2}.
2) Insert Constructed Affine Candidates
If the number of candidates in affine merge candidate list is less than MaxNumAffineCand (e.g., 5), constructed affine candidates are inserted into the candidate list. Constructed affine candidate means the candidate is constructed by combining the neighbor motion information of each control point.
If the number of candidates in affine merge candidate list is less than 5, zero motion vectors with zero reference indices are insert into the candidate list, until the list is full. More specifically, for the sub-block merge candidate list, a 4-parameter merge candidate with MVs set to (0, 0) and prediction direction set to uni-prediction from list 0 (for P slice) and bi-prediction (for B slice).
2.3.4. Merge with Motion Vector Differences (MMVD)
In JVET-L0054, ultimate motion vector expression (UMVE, also known as MMVD) is presented. UMVE is used for either skip or merge modes with a proposed motion vector expression method.
UMVE re-uses merge candidate as same as those included in the regular merge candidate list in VVC. Among the merge candidates, a base candidate can be selected, and is further expanded by the proposed motion vector expression method.
UMVE provides a new motion vector difference (MVD) representation method, in which a starting point, a motion magnitude and a motion direction are used to represent a MVD.
This proposed technique uses a merge candidate list as it is. But only candidates which are default merge type (MRG_TYPE_DEFAULT_N) are considered for UMVE's expansion.
Base candidate index defines the starting point. Base candidate index indicates the best candidate among candidates in the list as follows.
TABLE 1
Base candidate IDX
Base candidate
IDX
0
1
2
3
Nth MVP
1st MVP
2nd MVP
3rd MVP
4th MVP
If the number of base candidates is equal to 1, Base candidate IDX is not signaled.
Distance index is motion magnitude information. Distance index indicates the pre-defined distance from the starting point information. Pre-defined distance is as follows:
TABLE 2
Distance IDX
Distance
IDX
0
1
2
3
4
5
6
7
Pixel
1/4-pel
1/2-pel
1-pel
2-pel
4-pel
8-pel
16-pel
32-pel
distance
Direction index represents the direction of the MVD relative to the starting point. The direction index can represent of the four directions as shown below.
TABLE 3
Direction IDX
Direction IDX
00
01
10
11
x-axis
+
−
N/A
N/A
y-axis
N/A
N/A
+
−
UMVE flag is singnaled right after sending a skip flag or merge flag. If skip or merge flag is true, UMVE flag is parsed. If UMVE flage is equal to 1, UMVE syntaxes are parsed. But, if not 1, AFFINE flag is parsed. If AFFINE flag is equal to 1, that is AFFINE mode, But, if not 1, skip/merge index is parsed for VTM's skip/merge mode.
Additional line buffer due to UMVE candidates is not needed. Because a skip/merge candidate of software is directly used as a base candidate. Using input UMVE index, the supplement of MV is decided right before motion compensation. There is no need to hold long line buffer for this.
In current common test condition, either the first or the second merge candidate in the merge candidate list could be selected as the base candidate.
UMVE is also known as Merge with MV Differences (MMVD).
2.3.5. Decoder-Side Motion Vector Refinement (DMVR)
In bi-prediction operation, for the prediction of one block region, two prediction blocks, formed using a motion vector (MV) of list0 and a MV of list1, respectively, are combined to form a single prediction signal. In the decoder-side motion vector refinement (DMVR) method, the two motion vectors of the bi-prediction are further refined.
In JEM design, the motion vectors are refined by a bilateral template matching process. The bilateral template matching applied in the decoder to perform a distortion-based search between a bilateral template and the reconstruction samples in the reference pictures in order to obtain a refined MV without transmission of additional motion information. An example is depicted in
To further simplify the process of DMVR, JVET-M0147 proposed several changes to the design in JEM. More specifically, the adopted DMVR design to VTM-4.0 (to be released soon) has the following main features:
In JVET-L0100, multi-hypothesis prediction is proposed, wherein combined intra and inter prediction is one way to generate multiple hypotheses.
When the multi-hypothesis prediction is applied to improve intra mode, multi-hypothesis prediction combines one intra prediction and one merge indexed prediction. In a merge CU, one flag is signaled for merge mode to select an intra mode from an intra candidate list when the flag is true. For luma component, the intra candidate list is derived from 4 intra prediction modes including DC, planar, horizontal, and vertical modes, and the size of the intra candidate list can be 3 or 4 depending on the block shape. When the CU width is larger than the double of CU height, horizontal mode is exclusive of the intra mode list and when the CU height is larger than the double of CU width, vertical mode is removed from the intra mode list. One intra prediction mode selected by the intra mode index and one merge indexed prediction selected by the merge index are combined using weighted average. For chroma component, DM is always applied without extra signaling. The weights for combining predictions are described as follow. When DC or planar mode is selected, or the CB width or height is smaller than 4, equal weights are applied. For those CBs with CB width and height larger than or equal to 4, when horizontal/vertical mode is selected, one CB is first vertically/horizontally split into four equal-area regions. Each weight set, denoted as (w_intrai, w_interi), where i is from 1 to 4 and (w_intra1, w_inter1)=(6, 2), (w_intra2, w_inter2)=(5, 3), (w_intra3, w_inter3)=(3, 5), and (w_intra4, w_inter4)=(2, 6), will be applied to a corresponding region. (w_intra1, w_inter1) is for the region closest to the reference samples and (w_intra4, w_inter4) is for the region farthest away from the reference samples. Then, the combined prediction can be calculated by summing up the two weighted predictions and right-shifting 3 bits. Moreover, the intra prediction mode for the intra hypothesis of predictors can be saved for reference of the following neighboring CUs.
2.4 In-Loop Reshaping (ILR) in JVET-M0427
The in-loop reshaping (ILR) is also known as Luma Mapping with Chroma Scaling (LMCS).
The basic idea of in-loop reshaping (ILR) is to convert the original (in the first domain) signal (prediction/reconstruction signal) to a second domain (reshaped domain).
The in-loop luma reshaper is implemented as a pair of look-up tables (LUTs), but only one of the two LUTs need to be signaled as the other one can be computed from the signaled LUT. Each LUT is a one-dimensional, 10-bit, 1024-entry mapping table (1D-LUT). One LUT is a forward LUT, FwdLUT, that maps input luma code values Yi to altered values Yr: Yr=FwdLUT[Yi]. The other LUT is an inverse LUT, InvLUT, that maps altered code values Yr to Ŷi: Ŷi=InvLUT[Yr]. (Ŷi represents the reconstruction values of Yi.).
2.4.1 PWL Model
Conceptually, piece-wise linear (PWL) is implemented in the following way. Let x1, x2 be two input pivot points, and y1, y2 be their corresponding output pivot points for one piece. The output value y for any input value x between x1 and x2 can be interpolated by the following equation:
y=((y2−y1)/(x2−x1))*(x−x1)+y1
In fixed point implementation, the equation can be rewritten as:
y=((m*x+2FP_PREC−1)>>FP_PREC)+c
where m is scalar, c is an offset, and FP_PREC is a constant value to specify the precision.
In CE-12 software, the PWL model is used to precompute the 1024-entry FwdLUT and InvLUT mapping tables; but the PWL model also allows implementations to calculate identical mapping values on-the-fly without pre-computing the LUTs.
2.4.2. Test CE12-2
2.4.2.1. Luma Reshaping
Test 2 of the in-loop luma reshaping (i.e., CE12-2 in the proposal) provides a lower complexity pipeline that also eliminates decoding latency for block-wise intra prediction in inter slice reconstruction. Intra prediction is performed in reshaped domain for both inter and intra slices.
Intra prediction is always performed in reshaped domain regardless of slice type. With such arrangement, intra prediction can start immediately after previous TU reconstruction is done. Such arrangement can also provide a unified process for intra mode instead of being slice dependent.
CE12-2 also tests 16-piece piece-wise linear (PWL) models for luma and chroma residue scaling instead of the 32-piece PWL models of CE12-1.
Inter slice reconstruction with in-loop luma reshaper in CE12-2 (light-green shaded blocks indicate signal in reshaped domain: luma residue; intra luma predicted; and intra luma reconstructed)
2.4.2.2. Luma-Dependent Chroma Residue Scaling
Luma-dependent chroma residue scaling is a multiplicative process implemented with fixed-point integer operation. Chroma residue scaling compensates for luma signal interaction with the chroma signal. Chroma residue scaling is applied at the TU level. More specifically, the following applies:
The average is used to identify an index in a PWL model. The index identifies a scaling factor cScaleInv. The chroma residual is multiplied by that number.
It is noted that the chroma scaling factor is calculated from forward-mapped predicted luma values rather than reconstructed luma values
2.4.2.3. Signaling of ILR Side Information
The parameters are (currently) sent in the tile group header (similar to ALF). These reportedly take 40-100 bits.
The following tables are based on version 9 of JVET-L1001. The syntax to be added is highlighted below in underlined italicized font.
In 7.3.2.1 Sequence parameter set RBSP syntax:
Descriptor
seq_parameter_set_rbsp( ) {
sps_seq_parameter_set_id
ue(v)
. . .
sps_triangle_enabled_flag
u(1)
sps_ladf_enabled_flag
u(1)
if ( sps_ladf_enabled_flag ) {
sps_num_ladf_intervals_minus2
u(2)
sps_ladf_lowest_interval_qp_offset
se(v)
for( i = 0; i < sps_num_ladf_intervals_minus2 + 1; i++ ) {
sps_ladf_qp_offset[ i ]
se(v)
sps_ladf_delta_threshold_minus1[ i ]
ue(v)
}
}
sps_reshaper_enabled_flag
u(1)
rbsp_trailing_bits( )
}
In 7.3.3.1 General tile group header syntax:
Descriptor
tile_group_header( ) {
. . .
if( num_tiles_in_tile_group_minus1 > 0) {
offset_len_minus1
ue(v)
for( i = 0; i < num_tiles_in_tile_group_minus1; i++ )
entry_point_offset_minus1[ i ]
u(v)
}
if( sps_reshaper_enabled_flag ) {
tile group reshaper mode1_present flag
u(1)
if( tile_group reshaper model_present_flag)
tile_group_reshaper model ( )
tile_group reshaper_enable_flag
u(1)
if( tile_group_reshaper_enable_flag_&&(!( qtbtt_dual_tree_intra_flag &&
tile_group_type_= = I ) ) )
tile_group_reshaper_chro ma_residualscale_flag
u(1)
}
byte_alignment( )
}
A new syntax table tile group reshaper model can be added as follows:
Descriptor
tile_group_reshaper model_( ) {
reshaper model_min_bin_idx
ue(v)
reshaper_model_delta_max_bin_idx
ue(v)
reshaper_model_bin_delta_abs_cw_prec_minus1
ue(v)
for ( i = reshaper_model_min_bin_idx; i <= reshaper_model_max_bin_idx; i++ ) {
—
reshape_model_bin_delta_abs_CW [ i ]
u(v)
if( reshaper_model_bin_delta_abs_CW[ i ] ) > 0)
—
reshaper_model_bin_delta_sign_CW_flag[ i ]
u(1)
}
—
}
In General sequence parameter set RBSP semantics, the following semantics can be added:
sps_reshaper_enabled_flag equal to 1 specifies that reshaper is used in the coded video sequence (CVS). sps_reshaper_enabled_flag equal to 0 specifies that reshaper is not used in the CVS.
In tile group header syntax, the following semantics can be added:
tile_group_reshaper_model_present_flag equal to 1 specifies tile_group_reshaper_model( ) is present in tile group header. tile_group_reshaper_model_present_flag equal to 0 specifies tile_group_reshaper_model( ) is not present in tile group header. When tile_group_reshaper_model_present_flag is not present, it is inferred to be equal to 0.
tile_group_reshaper_enabled_flag equal to 1 specifies that reshaper is enabled for the current tile group. tile_group_reshaper_enabled_flag equal to 0 specifies that reshaper is not enabled for the current tile group. When tile_group_reshaper_enable_flag is not present, it is inferred to be equal to 0.
tile_group_reshaper_chroma_residual_scale_flag equal to 1 specifies that chroma residual scaling is enabled for the current tile group. tile_group_reshaper_chroma_residual_scale_flag equal to 0 specifies that chroma residual scaling is not enabled for the current tile group. When tile_group_reshaper_chroma_residual_scale_flag is not present, it is inferred to be equal to 0.
The tile_group_reshaper_model( ) syntax can be added as follows:
reshape_model_min_bin_idx specifies the minimum bin (or piece) index to be used in the reshaper construction process. The value of reshape_model_min_bin_idx shall be in the range of 0 to MaxBinIdx, inclusive. The value of MaxBinIdx shall be equal to 15.
reshape_model_delta_max_bin_idx specifies the maximum allowed bin (or piece) index MaxBinIdx minus the maximum bin index to be used in the reshaper construction process. The value of reshape_model_max_bin_idx is set equal to MaxBinIdx−reshape_model_delta_max_bin_idx.
reshaper_model_bin_delta_abs_cw_prec_minus1 plus 1 specifies the number of bits used for the representation of the syntax reshape_model_bin_delta_abs_CW[i].
reshape_model_bin_delta_abs_CW[i] specifies the absolute delta codeword value for the ith bin.
reshaper_model_bin_delta_sign_CW_flag[i] specifies the sign of reshape_model_bin_delta_abs_CW[i] as follows:
When reshape_model_bin_delta_sign_CW_flag[i] is not present, it is inferred to be equal to 0. The variable RspDeltaCW[i]=(1−2*reshape_model_bin_delta_sign_CW [i])*reshape_model_bin_delta_abs_CW [i].
The variable RspCW[i] is derived as following steps:
The value of RspCW [i] shall be in the range of 32 to 2*OrgCW−1 if the value of BitDepthY is equal to 10.
The variables InputPivot[i] with i in the range of 0 to MaxBinIdx+1, inclusive are derived as InputPivot[i]=i*OrgCW.
The variable ReshapePivot[i] with i in the range of 0 to MaxBinIdx+1, inclusive, the variables ScaleCoef[i] and InvScaleCoeff[i] with i in the range of 0 to MaxBinIdx, inclusive, can be derived as follows:
shiftY = 14
ReshapePivot[ 0 ] = 0;
for( i = 0; i <= MaxBinIdx ; i++) {
ReshapePivot[ i + 1] = ReshapePivot[ i ] + RspCW[ i ]
ScaleCoef[ i ] = ( RspCW[ i ] * (1 << shiftY) +
(1 << (Log2(OrgCW) − 1))) >> (Log2(OrgCW))
if ( RspCW[ i ] = = 0)
InvScaleCoeff[ i ] = 0
else
InvScaleCoef[ i ] = OrgCW * (1 << shiftY) / RspCW[ i ]
}
The variable ChromaScaleCoef[i] with i in the range of 0 to MaxBinIdx, inclusive, can be derived as follows:
At the encoder side, each picture (or tile group) is firstly converted to the reshaped domain. And all the coding process is performed in the reshaped domain. For intra prediction, the neighboring block is in the reshaped domain; for inter prediction, the reference blocks (generated from the original domain from decoded picture buffer) are firstly converted to the reshaped domain. Then the residual(s) is/are generated and coded to the bitstream.
After the whole picture (or tile group) finishes encoding/decoding, samples in the reshaped domain are converted to the original domain, then deblocking filter and other filters are applied.
Forward reshaping to the prediction signal is disabled for the following cases. The prediction signal, for example, comprises a prediction block of same size as the current block.
Virtual pipeline data units (VPDUs) are defined as non-overlapping M×M-luma(L)/N×N-chroma(C) units in a picture. In hardware decoders, successive VPDUs are processed by multiple pipeline stages at the same time; different stages process different VPDUs simultaneously. The VPDU size is roughly proportional to the buffer size in most pipeline stages, so it is said to be very important to keep the VPDU size small. In HEVC hardware decoders, the VPDU size is set to the maximum transform block (TB) size. Enlarging the maximum TB size from 32×32-L/16×16-C (as in HEVC) to 64×64-L/32×32-C (as in the current VVC) can bring coding gains, which results in 4× of VPDU size (64×64-L/32×32-C) expectedly in comparison with HEVC. However, in addition to quadtree (QT) coding unit (CU) partitioning, ternary tree (TT) and binary tree (BT) are adopted in VVC for achieving additional coding gains, and TT and BT splits can be applied to 128×128-L/64×64-C coding tree blocks (CTUs) recursively, which is said to lead to 16× of VPDU size (128×128-L/64×64-C) in comparison with HEVC.
In current design of VVC, the VPDU size is defined as 64×64-L/32×32-C.
2.6. APS
An Adaptation Parameter Set (APS) is adopted in VVC to carry ALF parameters. The tile group header contains an aps_id which is conditionally present when ALF is enabled. The APS contains an aps_id and the ALF parameters. A new NUT (NAL unit type, as in AVC and HEVC) value is assigned for APS (from JVET-M0132). For the common test conditions in VTM-4.0 (to appear), it is suggested just using aps_id=0 and sending the APS with each picture. For now, the range of APS ID values will be 0 . . . 31 and APSs can be shared across pictures (and can be different in different tile groups within a picture). The ID value should be fixed-length coded when present. ID values cannot be re-used with different content within the same picture.
2.7. Post-Reconstruction Filters
2.7.1 Diffusion Filter (DF)
In JVET-L0157, diffusion filter is proposed, wherein the intra/inter prediction signal of the CU may be further modified by diffusion filters.
2.7.1.1. Uniform Diffusion Filter
The Uniform Diffusion Filter is realized by convolving the prediction signal with a fixed mask that is either given as hI or as hIV, defined below.
Besides the prediction signal itself, one line of reconstructed samples left and above of the block are used as an input for the filtered signal, where the use of these reconstructed samples can be avoided on inter blocks.
Let pred be the prediction signal on a given block obtained by intra or motion compensated prediction. In order to handle boundary points for the filters, the prediction signal needs to be extended to a prediction signal predext. This extended prediction can be formed in two ways: Either, as an intermediate step, one line of reconstructed samples left and above the block are added to the prediction signal and then the resulting signal is mirrored in all directions. Or only the prediction signal itself is mirrored in all directions. The latter extension is used for inter blocks. In this case, only the prediction signal itself comprises the input for the extended prediction signal predext.
If the filter hI is to be used, it is proposed to replace the prediction signal pred by
hI*pred,
using the aforementioned boundary extension. Here, the filter mask hI is given as
If the filter hIV is to be used, it is proposed to replace the prediction signal pred by
hIV*pred.
Here, the filter hIV is given as hIV=hI*hI*hI*hI.
2.7.1.2. Directional Diffusion Filter
Instead of using signal adaptive diffusion filters, directional filters, a horizontal filter hhor and a vertical filter hver, are used which still have a fixed mask. More precisely, the uniform diffusion filtering corresponding to the mask hI of the previous section is simply restricted to be either applied only along the vertical or along the horizontal direction. The vertical filter is realized by applying the fixed filter mask
to the prediction signal and the horizontal filter is realized by using the transposed mask hhor=hvert.
2.7.2. Bilateral Filter (BF)
Bilateral filter is proposed in JVET-L0406, and it is always applied to luma blocks with non-zero transform coefficients and slice quantization parameter larger than 17. Therefore, there is no need to signal the usage of the bilateral filter. Bilateral filter, if applied, is performed on decoded samples right after the inverse transform. In addition, the filter parameters, i.e., weights are explicitly derived from the coded information.
The filtering process is defined as:
P′0,0=P0,0+Σk=1KWk(abs(Pk,0−P0,0))×(Pk,0−P0,0), (1)
where P0,0 is the intensity of the current sample and P′0,0 is the modified intensity of the current sample, Pk,0 and Wk are the intensity and weighting parameter for the k-th neighboring sample, respectively. An example of one current sample and its four neighboring samples (i.e., K=4) is depicted in
More specifically, the weight Wk(x) associated with the k-th neighboring sample is defined as follows:
and σd is dependent on the coded mode and coding block sizes. The described filtering process is applied to intra-coded blocks, and inter-coded blocks when TU is further split, to enable parallel processing.
To better capture statistical properties of video signal, and improve performance of the filter, weights function resulted from Equation (2) are being adjusted by the σd parameter, tabulated in Table 4 as being dependent on coding mode and parameters of block partitioning (minimal size).
TABLE 4
Value of σd for different block sizes and coding modes
Min (block width,
block height)
Intra mode
Inter mode
4
82
62
8
72
52
Other
52
32
To further improve the coding performance, for inter-coded blocks when TU is not split, the intensity difference between current sample and one of its neighboring samples is replaced by a representative intensity difference between two windows covering current sample and the neighboring sample. Therefore, the equation of filtering process is revised to:
wherein Pk,m and P0,m represent the m-th sample value within the windows centered at Pk,0 and P0,0, respectively. In this proposal, the window size is set to 3=3. An example of two windows covering P2,0 and P0,0 are depicted in
2.7.3. Hadamard Transform Domain Filter (HF)
In JVET-K0068, in-loop filter in 1D Hadamard transform domain which is applied on CU level after reconstruction and has multiplication free implementation. Proposed filter is applied for all CU blocks that meet the predefined condition and filter parameters are derived from the coded information.
Proposed filtering is always applied to luma reconstructed blocks with non-zero transform coefficients, excluding 4×4 blocks and if slice quantization parameter is larger than 17. The filter parameters are explicitly derived from the coded information. Proposed filter, if applied, is performed on decoded samples right after inverse transform.
For each pixel from reconstructed block pixel processing comprises the following steps:
wherein (i) is index of spectrum component in Hadamard spectrum, R(i) is spectrum component of reconstructed pixels corresponding to index, σ is filtering parameter deriving from codec quantization parameter QP using following equation:
σ=2(1+0.126*(QP−27))
The example of scan pattern is depicted on
2.8 Clipping Functions
During the prediction stage (either intra or inter or combined intra-inter), reconstruction stage before in-loop filtering and during in-loop filtering, the predicted/reconstructed/filtered samples need to be clipped to a range.
Some clipping functions are defined in the standard, such as:
In equation (3), y indicates the minimum value of the clipping parameter, and x indicates the maximum value of the clipping parameter.
Just to give some examples for the usage of clipping functions, some sub-clause(s) in the VVC specification (JVET-M1001 version 5) are listed herein.
2.8.1 Clipping in the Intra Prediction Stage
8.4.4.2.7 Specification of INTRA_ANGULAR2 . . . INTRA_ANGULAR66 Intra Prediction Modes
Inputs to this process are:
Outputs of this process are the modified intra prediction mode predModeIntra and the predicted samples predSamples[x][y], with x=0 . . . nTbW−1, y=0 . . . nTbH−1.
. . .
The values of the prediction samples predSamples[x][y], with x=0 . . . nTbW−1, y=0 . . . nTbH−1 are derived as follows:
The current design of ILR may have the following problems:
The detailed embodiments described below should be considered as examples to explain general concepts. These embodiments should not be interpreted narrowly way. Furthermore, these embodiments can be combined in any manner.
Signaling of ILR Side Information
if ( sps_reshaper_enabled_flag ) {
tile_group_reshaper_model_present_flag
u(1)
if ( tile_group_reshaper_model_present_flag )
tile_group_reshaper_model ( )
if ( tile_group_type != I ) {
if( tile_group_type = = B )
reshaper_model_from_l1 flag
u(1)
reshaper_mode_from_refidx
}
tile_group_reshaper_enable_flag
u(1)
if ( tile_group_reshaper_enable_flag &&
(!( qtbtt_dual_tree_intra_flag && tile_group_type == I ) ) )
tile_group_reshaper_chr oma_residual_scale_flag
u(1)
}
from a reference picture in reference list 1. Otherwise, the reshaping information is inherited from a reference picture in reference list 0. When not present, it can be inferred to be 0.
In this section, some example embodiments are described with reference to a bitstream syntax that shows how the bitstream syntax of the current version of VVC can be modified based on some techniques described in the present document. Various syntax element in the description below have meanings similar to meanings described in the VVC document. Furthermore, where applicable, new syntax elements and their descriptions are highlighted using bold face italicized text.
Syntax design in one embodiment is shown as below. General tile group header syntax
Descriptor
tile_group_header( ) {
tile_group_pic_parameter_set_id
ue(v)
if( NumTilesInPic > 1 ) {
tile_group_address
u(v)
num_tiles_in_tile_group_minus1
ue(v)
}
tile_group_type
ue(v)
tile_group_pic_order_cnt_lsb
u(v)
if( partition_constraints_override_enabled_flag ) {
partition_constraints_override_flag
ue(v)
if( partition_constraints_override_flag ) {
tile_group_log2_diff_min_qt_min_cb_luma
ue(v)
tile_group_max_mtt_hierarchy_depth_luma
ue(v)
if( tile_group_max_mtt_hierarchy_depth_luma_!= 0)
tile_group_log2_diff_max_bt_min_qt_luma
ue(v)
tile_group_log2_diff_max_tt_min_qt_luma
ue(v)
}
if( tile_group_type = = I && qtbtt_dual_tree_intra_flag ) {
tile_group_log2_diff_min_qt_min_cb_chroma
ue(v)
tile_group_max_mtt_hierarchy_depth_chroma
ue(v)
if( tile_group_max_mtt_hierarchy_depth_chroma != 0)
tile_group_log2_diff_max_bt_min_qt_chroma
ue(v)
tile_group_1og2_diff_max_tt_min_qt_chroma
ue(v)
}
}
}
}
if ( tile_group_type != I)
if( sps_temporal_mvp_enabled_flag )
tile_group_temporal_mvp_enabled_flag
u(1)
if( tile_group_type = = B)
mvd_l1_zero_flag
u(1)
if( tile_group_temporal_mvp_enabled_flag ) {
if( tile_group_type = = B)
collocated_from_l0_flag
u(1)
}
six_minus_max_num_merge_cand
ue(v)
if( sps_affine_enable_flag )
five_minus_max_num_subblock_merge_cand
ue(v)
}
tile_group_qp_delta
se(v)
if( pps_tile_group_chroma_qp_offsets_present_flag ) {
tile_group_cb_qp_offset
se(v)
tile_group_cr_qp_offset
se(v)
}
if( sps_sao_enabled_flag ) {
tile_group_sao_luma_flag
u(1)
if( ChromaArrayType != 0)
tile_group_sao_chroma_flag
u(1)
}
if( sps_alf_enabled_flag ) {
tile_group_alf_enabled_flag
u(1)
if( tile_group_alf_enabled_flag )
alf_data( )
}
if( tile_group_type == P | | tile_group_type = = B) {
num_ref_idx_l0_active_minus1
ue(v)
if( tile_group_type = =B )
num_ref_idx_l1_active_minus1
ue(v)
}
dep_quant_enabled_flag
u(1)
if( !dep_quant_enabled_flag )
sign_data_hiding_enabled_flag
u(1)
if( deblocking_filter_override_enabled_flag )
deblocking_filter_override_flag
u(1)
if( deblocking_filter_override_flag ) {
tile_group_deblocking_filter_disabled_flag
u(1)
if( !tile_group_deblocking_filter_disabled_flag ) {
tile_group_beta_offset_div2
se(v)
tile_group_tc_offset_div2
se(v)
}
}
if( num_tiles_in_tile_group_minus1 > 0 ) {
offset_len_minus1
ue(v)
for( i = 0; i < num_tiles_in_tile_group_minus1; i++ )
entry_point_offset_minus1[ i ]
u(v)
}
if ( sps_reshaper_enabled_flag ) {
tile_group_reshaper_model_present_flag
u(1)
if ( tile_group_reshaper_model_present_flag )
tile_group_reshaper_model ( )
else if ( tile_group_type != I ) {
if( tile_group_type = = B)
reshaper_model_from_l1_flag
u(1)
reshapter_mode_from_refidx
ue(v)
}
tile_group_reshaper_enable_flag
u(1)
if ( tile_group_reshaper_enable_flag && (!( qtbtt_dual_tree_intra_flag &&
tile_group_type == I) ) )
tile_group_reshaper_chroma_residual_scale_flag
u(1)
}
byte_alignment( )
}
Add a new syntax table tile group reshaper model:
Descriptor
tile_group_reshaper_model ( ) {
reshaper_model_min_bin_idx
ue(v)
reshaper_model_delta_max_bin_idx
ue(v)
reshaper_model_bin_delta_abs_cw_prec_minus1
ue(v)
for ( i = reshaper_model_min_bin_idx; i <= reshaper_model_max_bin_idx; i++ ) {
reshape_model_bin_delta_abs_CW [ i ]
u(v)
if ( reshaper_model_bin_delta_abs_CW[ i ] ) > 0 )
reshaper_model_bin_delta_sign_CW_flag[ i ]
u(1)
}
}
shiftY = 14
ReshapePivot[ 0 ] = 0;
for( i = 0; i <= MaxBinIdx ; i++) {
if RspCW[ i ] = = 0
ScaleCoef[ i ] = InvScaleCoeff[ i ]= 1 << shiftY,
else
ReshapePivot[ i +1 ] = ReshapePivot[ i ] + RspCW[ i ]
ScaleCoef[ i ] = ( RspCW[ i ] *
(1 << shiftY) +(1 << (Log2(OrgCW) − 1))) >>
(Log2(OrgCW))
if ( RspCW[ i ] = = 0 )
InvScaleCoeff [ i ] = 0
else
InvScaleCoeff[ i ] = OrgCW * (1 << shiftY) / RspCW[ i ]
}
The system 4100 may include a coding component 4104 that may implement the various coding or encoding methods described in the present document. The coding component 4104 may reduce the average bitrate of video from the input 4102 to the output of the coding component 4104 to produce a coded representation of the video. The coding techniques are therefore sometimes called video compression or video transcoding techniques. The output of the coding component 4104 may be either stored, or transmitted via a communication connected, as represented by the component 4106. The stored or communicated bitstream (or coded) representation of the video received at the input 4102 may be used by the component 4108 for generating pixel values or displayable video that is sent to a display interface 4110. The process of generating user-viewable video from the bitstream representation is sometimes called video decompression. Furthermore, while certain video processing operations are referred to as “coding” operations or tools, it will be appreciated that the coding tools or operations are used at an encoder and corresponding decoding tools or operations that reverse the results of the coding will be performed by a decoder.
Examples of a peripheral bus interface or a display interface may include universal serial bus (USB) or high definition multimedia interface (HDMI) or Displayport, and so on. Examples of storage interfaces include SATA (serial advanced technology attachment), PCI, IDE interface, and the like. The techniques described in the present document may be embodied in various electronic devices such as mobile phones, laptops, smartphones or other devices that are capable of performing digital data processing and/or video display.
In the present document, the term “video processing” may refer to video encoding, video decoding, video compression or video decompression. For example, video compression algorithms may be applied during conversion from pixel representation of a video to a corresponding bitstream representation or vice versa. The bitstream representation of a current video block may, for example, correspond to bits that are either co-located or spread in different places within the bitstream, as is defined by the syntax. For example, a macroblock may be encoded in terms of transformed and coded error residual values and also using bits in headers and other fields in the bitstream.
It will be appreciated that the disclosed methods and techniques will benefit video encoder and/or decoder embodiments incorporated within video processing devices such as smartphones, laptops, desktops, and similar devices by allowing the use of the techniques disclosed in the present document.
In some implementations, the side information applicable to the some of the one or more video regions is coded without directly inheriting from other video regions. In some implementations, the coded representation includes the side information at a video region level that includes a current video block. In some implementations, a clipping process is used for the conversion based on clipping parameters that are in the first domain and/or in the second domain according to the ILR information. In some implementations, the coded representation includes information applicable for in-loop reshaping that provides ILR parameters including clipping parameters used according to a rule for a clipping process during the ILR. In some implementations, the conversion is performed between a current video block of the video that is a chroma block and the coded representation of the video. In some implementations, the conversion further includes applying a reshaping process to one or more chroma components of the current video block based on the side information. In some implementations, the side information for a current video region is determined according to a rule that disallows inheritance or prediction of the side information based on a temporal layer index of the current video region. In some implementations, the side information for a current video region is determined according to a rule that disallows inheritance or prediction of the side information from a first type of another picture.
In some implementations, the conversion is performed between a current video block of the video and the coded representation of the video using an in loop reshaping process. In some implementations, the in loop reshaping includes obtaining a prediction block from motion compensation prediction or an intra prediction; applying a forward reshaping to the prediction block to generate a forward reshaped prediction block in a first domain; reconstructing, based on the forward reshaped prediction block, the coded representation of the video to generate a first reconstruction block in a second domain; applying an inverse reshaping to the first reconstruction block to obtain an inverse reconstruction block; and applying a loop filter to the inverse reconstruction block to obtain a final reconstruction block. In some implementations, a single clipping process is used during the in loop reshaping process.
In some implementations, the coded representation includes a first side information at a first level, wherein a second side information at a second level is derived from the first side information such that the second side information provides parameters for a video unit coded with in-loop reshaping (ILR) in which a reconstruction of the video unit of a video region is based on a representation of a video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit.
Various techniques and embodiments may be described using the following clause-based format.
The first set of clauses describe certain features and aspects of the disclosed techniques listed in the previous section.
1. A method of visual media processing, comprising:
2. The method of clause 1, wherein the side information associated with the in-loop reshaping step includes one or more lookup tables or parameters for piece-wise linear models related to chroma samples or luma samples.
3. The method of clause 2, wherein differences between the current video block and another video block are predictively coded in the side information, wherein the differences are related to at least one parameter in the parameters for piece-wise linear models.
4. The method of clause 2, wherein indices of the side information are signaled in the bitstream information.
5. The method of clause 2, wherein the side information are subsets of the in-loop reshaping step.
6. The method of clause 2, wherein the indices of the side information are included in a tile group header, slice header, or picture header.
7. The method of any one or more of clauses 1-6, wherein the side information is predicted from information stored in a first-in-first-out buffer.
8. The method of clause 7, wherein the buffer is updated based on the parameters for piece-wise linear models in connection with the current video block.
9. The method of clause 7, wherein the buffer stores the side information for a first video block, further comprising:
10. The method of clause 7, wherein the buffer stores the side information for a first video block, further comprising:
11. A method of visual media processing, comprising:
12. The method of clause 11, wherein the inheritance or the prediction is enabled when a temporal layer index of the reference picture is no larger than a temporal index layer for the current video block.
13. The method of clause 14, wherein the reference picture is stored in a decoded picture buffer.
14. The method of clause 11, wherein the inheritance or the prediction is based on a list or an index.
15. The method of clauses 11-14, wherein the reference picture is predefined.
16. The method of clauses 15, wherein the reference picture is a collocated picture.
17. The method of clause 14, wherein the list is denoted as LX and the index is denoted as IDX, and wherein LX=0 and IDX=0.
18. The method of any one or more of clauses 15-16, wherein the reference picture is closest in decoding order to the current video block.
19. The method of clause 11, wherein the another video block is in a decoded picture buffer.
20. A method of visual media processing, comprising:
21. The method of 20, wherein the side information is included in a header associated with a group of tiles, and wherein the header is accessible by the first tile and the second tile.
22. The method of clause 20, wherein the first tile is a group of tiles and the second tile is a singular tile.
23. The method of clause 20, wherein the first tile is a picture and the second tile is a group of tiles.
24. The method of clause 20, wherein the first tile is an adaptation parameter set (APS) and the second tile is a group of tiles.
25. The method of clause 20, wherein the first tile is an adaptation parameter set (APS) and the second tile is a picture.
26. The method of clause 20, wherein the first tile is an adaptation parameter set (APS) and the second tile is a singular tile.
27. A method of visual media processing, comprising:
28. The method of clause 27, wherein the side information is set to a default value.
29. The method of clause 27, wherein the default value is set as forwardLUT(x), wherein forwardLUT(.) is a lookuptable and x is a value used when the in-loop reshaping step is disabled.
30. The method of clause 29, wherein the value x is set as (1<<(Bitdepth−1)), wherein Bitdepth is an input bitdepth value of the current video block or a video block reconstructed from the current video block.
31. A method of visual media processing, comprising:
32. The method of clause 31, wherein a minimum value of the clipping parameters is defined when x=0.
33. The method of clause 31, wherein a maximum value of the clipping parameters is defined when x=(1<<(Bitdepth−1)), wherein Bitdepth is an input bitdepth value of the current video block or a video block reconstructed from the current video block.
34. The method of clause 31, wherein a minimum value of the parameters is 0.
35. The method of clause 31, wherein a minimum value of the parameters is defined as inverseLUT(x), wherein x is the largest value obtained by solving InverseLUT(x) unequal to InverseLUT(x−1).
36. The method of clause 35, wherein x lies in a range [0, maximum entry index].
37. The method of clause 35, wherein x lies in a range [0, (1<<(Bitdepth−1))], wherein Bitdepth is an input bitdepth value of the current video block or a video block reconstructed from the current video block.
38. The method of clause 31, wherein the parameters are used in at least a portion of one or more of the following steps: intra and/or intra block copy prediction block generation, palette mode, motion compensation, inter prediction block generation, final reconstruction block derivation, filtering.
39. The method of any one or more of clauses 31-38, further comprising:
40. The method of any one or more of clauses 31-38, further comprising:
41. The method of any one or more of clauses 1-40, further comprising:
42. The method of clause 41, wherein the condition is related to any one or more of: a luma color component, a main color component, a color format, or a chroma component.
43. The method of clause 31, further comprising:
44. The method of any one or more of clauses 1-43, wherein the side information is included in a tile group header, a slice header, or a picture header.
45. A method of visual media processing, comprising:
46. The method of clause 45, wherein the reference picture is associated with a first reference list in the one or more reference lists, and wherein the side information includes a flag with a value one.
47. The method of clause 46, wherein the first reference list is denoted as reference list 0.
48. The method of clause 46, wherein the first reference list is denoted as reference list 1.
49. The method of clause 45, wherein, if the another video block is the non-reference picture, the inheritance or the prediction is disabled, then the side information includes a flag with a value zero.
50. The method of clause 45, wherein, if (a) a flag denoted tile_group_reshaper_model_present_flag and present in the tile group header, the slice header, or the picture header equals 0, and (b) the tile group header, the slice header, or the picture header is of I-type, then the side information is initialized in accordance with an initialization step.
51. The method of clause 45, wherein, if the another video block is a reference picture, the inheritance or the prediction is enabled from the side information of a first tile group or a first slice of the reference picture.
52. The method of clause 51, wherein the side information of the another video block is a first side information, the side information of the current video block is a second side information, and wherein the second side information is inherited from the first side information.
53. The method of clause 52, wherein the second side information inheriting from the first side information includes copying from the first side information one or more of the following quantitates into associated one or more quantities in the second side information: reshaper_model_min_bin_idx, reshaper_model_max_bin_idx, RspCW[i], RspDeltaCW [i], ReshapePivot[i], ScaleCoef [i], InvScaleCoeff [i], ChromaScaleCoef [i], tile_group_reshaper_enable_flag, or tile_group_reshaper_chroma_residual_scale_flag, for i ranging between a lower limit and an upper limit.
54. The method of clause 53, wherein the lower limit and the upper limit respectively are reshaper_model_min_bin_idx and reshaper_model_max_bin_idx.
55. The method of clause 53, wherein the lower limit and the upper limit respectively are zero and MaxBinIdx.
56. The method of clause 45, wherein the reference picture is selected in response to determining that a tile group header, a slice header, or a picture header of the reference picture is of I-type.
57. The method of clause 45, wherein the reference picture is selected from a group of reference pictures in response to determining that a tile group header, a slice header, or a picture header of the group of reference pictures is of I-type.
58. The method of clause 57, wherein the reference picture is closest to the current video block.
59. The method of clause 58, wherein an absolute difference of a picture order count (POC) of the reference picture and the current video block is smallest in the group of reference pictures.
60. The method of clause 45, wherein, in response to determining that a tile group header, a slice header, or a picture header of the reference picture lacks I-type, further comprising:
61. The method of any one or more of clauses 1-60, wherein the visual media processing is an encoder-side implementation.
62. The method of any one or more of clauses 1-60, wherein the visual media processing is a decoder-side implementation.
63. An apparatus in a video system comprising a processor and a non-transitory memory with instructions thereon, wherein the instructions upon execution by the processor, cause the processor to implement the method in any one or more of clauses 1-60.
64. A computer program product stored on a non-transitory computer readable media, the computer program product including program code for carrying out the method in any one or more of clauses 1-60.
The second set of clauses describe certain features and aspects of the disclosed techniques listed in the previous section, including, for example, Example Implementations 1, 2, 4, 6-13.
1. A method of video coding, comprising: performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes side information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the side information provides parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the side information applicable to the some of the one or more video regions is coded without directly inheriting from other video regions.
2. The method of clause 1, wherein the side information includes one or more lookup tables or parameters for one or more piece-wise linear models related to chroma samples or luma samples.
3. The method of clause 1, wherein the video region is a picture, slice, tile group, or other video data unit.
4. The method of clause 1, wherein the side information is predictively coded in the coded representation.
5. The method of clause 1, wherein differences between the side information of the video unit and the side information of another video unit are predictively coded, wherein the differences are related to at least one parameter in parameters for a piece-wise linear model.
6. The method of clause 1, wherein only some of parameters of a piece-wise linear model is inherited and remaining parameters are signaled.
7. The method of clause 2, wherein the side information is coded by using one or multiple sets of side information corresponding to one or multiple piece-wise linear models.
8. The method of clause 2, wherein one or multiple indices of the multiple sets of piece-wise linear models are signaled in the coded representation of the video.
9. The method of clause 8, wherein the side information that has been signaled is used to derive the side information of the video unit.
10. The method of clause 8, wherein the one or multiple indices of the side information are included in a tile group header, slice header, or picture header.
11. The method of clause 1, wherein one or multiple sets of the side information are signaled in an adaptation parameter set (APS).
12. The method of clause 1, wherein one or multiple sets of the side information are stored in a buffer.
13. The method of clause 1, wherein parameters coded in the coded representation are utilized to update the buffer.
14. The method of clause 1, wherein the side information for the video region is inherited from one of multiple sets of the side information stored in the buffer.
15. The method of clause 14, wherein an index of the one of multiple sets of side information is signaled.
16. The method of clause 1, wherein the side information for the video region is predicted from one or multiple sets of side information stored in the buffer.
17. The method of clause 16, wherein an index of the one or multiple sets of the side information is signaled.
18. A method for video processing, comprising: performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes side information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the side information provides parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the coded representation includes the side information at a video region level that includes a current video block.
19. The method of clause 18, wherein the video region level includes a tile level or a coding tree unit (CTU) level.
20. The method of clause 18, wherein, for a picture or a tile group including multiple tiles, an inheritance or prediction of the side information of the current video block between tiles is allowed.
21. The method of clause 18, wherein, for a picture or a tile group including multiple tiles, an inheritance or prediction of the side information of the current video block is disallowed.
22. The method of clause 18, wherein the video region level is a tile group header and all tiles within a corresponding tile group share the side information.
23. A video processing method, comprising: determining a prediction block of a current video block, for a conversion between the current video block of a video region of a video and a coded representation of the video, using samples from the video region, wherein unavailable samples of the samples from the video region are determined based on a derivation rule that specifies to use an in-loop reshaping (ILR) information associated with the current video block; and performing the conversion based on the prediction block, wherein the in-loop reshaping information includes information for a reconstruction of a video unit of the video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit.
24. The method of clause 23, wherein, for the current video block coded in intra coding or coded in a cross-component linear model (CCLM) mode, a default value is set for the in-loop reshaping information.
25. The method of clause 24, wherein the default value is set as forwardLUT(x), wherein forwardLUT(.) is a lookuptable and x is a value used when the coding mode is disabled.
26. The method of clause 25, wherein the value x is set as (1<<(Bitdepth−1)), wherein Bitdepth is an input bitdepth value of the current video block or a video block reconstructed from the current video block.
27. A video processing method, comprising: performing a conversion between a current video block of a video and a coded representation of the video using an in loop reshaping process, wherein the in loop reshaping process includes: obtaining a prediction block from motion compensation prediction or an intra prediction; applying a forward reshaping to the prediction block to generate a forward reshaped prediction block in a first domain; reconstructing, based on the forward reshaped prediction block, the coded representation of the video to generate a first reconstruction block in a second domain; applying an inverse reshaping to the first reconstruction block to obtain an inverse reconstruction block; applying a loop filter to the inverse reconstruction block to obtain a final reconstruction block, and wherein a single clipping process is used during the in loop reshaping process.
28. The method of clause 27, wherein the single clipping process is used with clipping parameters defined in the first domain and derived from a look up table or a piece-wise linear (PWL) model used in the in loop reshaping process.
29. The method of clause 28, wherein a minimum value of the clipping parameters is defined as inverseLUT(x), wherein inverseLUT(.) is an inverse lookup table and x is 0.
30. The method of clause 28, wherein the clipping parameters is defined as inverseLUT(y), wherein inverseLUT(.) is an inverse lookup table and y is unequal to 0.
31. The method of clause 27, wherein the single clipping process is used with clipping parameters defined in the second domain and derived from a look up table or a piece-wise linear (PWL) model used in the in loop reshaping process.
32. The method of clause 31, wherein a minimum value of the clipping parameters is defined as a default value that is 0.
33. The method of clause 27, wherein a minimum value of the clipping parameters is defined as inverseLUT(y), wherein inverseLUT(.) is an inverse lookup table and y is the largest value obtained by solving InverseLUT(y) unequal to InverseLUT(y−1) or InverseLUT(y+1).
34. The method of clause 33, wherein y lies in a range [0, maximum entry index] or in a range [0, (1<<(Bitdepth−1))], wherein Bitdepth is an input bitdepth value of the current video block or a video block reconstructed from the current video block.
35. A video processing method, comprising: performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes in-loop reshaping (ILR) information applicable for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein a clipping process is used for the conversion based on clipping parameters that are in the first domain and/or in the second domain according to the ILR information.
36. The method of clause 35, wherein the clipping parameters are used during a prediction block generation process, the prediction block generation process including at least one of intra and/or intra block copy generation or inter prediction block generation.
37. The method of clause 35, wherein the clipping parameters are used during the conversion performed in a palette mode in which at least a palette of representative sample values is used for the video unit.
38. The method of clause 35, wherein the clipping parameters are used during a motion compensation process.
39. The method of clause 35, wherein the clipping parameters are used during a final reconstruction block derivation process.
40. The method of clause 35, wherein the clipping parameters are used during a filtering process.
41. A video processing method, comprising: performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the information provides ILR parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the ILR parameters include clipping parameters used according to a rule for a clipping process during the ILR.
42. The method of clause 41, wherein the rule defines different values for the clipping parameters at different stages of ILR.
43. The method of clause 42, wherein the clipping parameters are defined in the second domain and used for the conversion before an ILR filtering operation.
44. The method of clause 42, wherein the clipping parameters are defined in the second domain and used for a prediction block process and/or a reconstruction of the video unit.
45. The method of clause 42, wherein the clipping parameters are defined in the first domain and used during an in-loop filtering operation applied to the video unit.
46. The method of clause 43 or 45, wherein the in-loop filtering operation is performed using a deblocking filter, a sample adaptive offset filter, or an adaptive loop filter.
47. A video processing method, comprising: determining whether an in loop reshaping (ILR) is enabled for a conversion between a coded representation of a video comprising one or more video regions and the video; and performing the conversion based on the determining, wherein a clipping process is used for the conversion based on ILR clipping parameters derived according to information associated with the ILR and/or fixed clipping parameters unrelated to the information associated with the ILR, and wherein the information associated with the ILR is applicable for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit.
48. The method of clause 47, wherein the determining determines that the ILR is disabled, and wherein the fixed clipping parameters are used for the conversion.
49. The method of clause 47, wherein the determining determines that the ILR is enabled, and wherein the ILR clipping parameters represented in the first domain and/or the second domain are used for the conversion.
50. The method of clause 47, wherein the determining further determined whether an adaptive clipping process is enabled for the conversion, and the clipping process is used for the conversion based on at least one of ILR clipping parameters, fixed clipping parameters, or adaptive clipping parameters.
51. The method of clause 50, wherein the determining determines that the ILR is disabled and that the adaptive clipping process is enabled, and the adaptive clipping parameters are used for the conversion.
52. The method of clause 50, wherein the determining determines that the ILR is enabled and the ILR clipping parameters represented in the first domain and/or the second domain are used for the conversion.
53. The method of clause 50, wherein the determining determines that the ILR is disabled and that the adaptive clipping is disabled, and the fixed clipping parameters are used for the conversion.
54. The method of any one of clauses 1 to 53, further comprising detecting a certain condition associated with the video unit.
55. The method of clause 54, wherein the certain condition is whether the coding mode is enabled or disabled for a video region including the video unit.
56. The method of clause 54, wherein the condition is related to any one or more of: a luma color component, a main color component, a color format, or a chroma component.
57. A method for video processing, comprising: performing a conversion between a current video block of a video that is a chroma block and a coded representation of the video, wherein the coded representation includes side information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the side information provides parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the conversion further includes applying a reshaping process to one or more chroma components of the current video block based on the side information.
58. The method of clause 57, wherein the side information associated with the coding mode includes look up tables and/or piece-wise linear (PWL) models.
59. The method of clause 57, wherein the method further includes: determining whether a color format of the current video block is 4:4:4; and refraining from applying a luma-based residual scaling to the one or more chroma components of the current video block based on the determining.
60. The method of clause 59, wherein the color format of the current video block is 4:4:4, and wherein the luma-based residual scaling is replaced by the look up tables and/or piece-wise linear models that are signaled or derived.
61. The method of clause 57, wherein two chroma components of the current video block share the side information associated with the coding mode.
62. The method of clause 57, wherein the method further includes: determining whether a color format of the current video block is 4:4:4; and deriving look up tables and/or piece-wise linear models for chroma components from those for the luma component.
63. The method of clause 57, wherein the method further includes: determining whether a color format of the current video block is 4:4:4, and wherein look up tables and/or piece-wise linear models for chroma components are signaled in an adaptation parameter set (APS), a picture parameter set (PPS), a picture header, a slice header, a tile group header, a tile, or other video units.
64. The method of clause 57, wherein the method further includes: determining whether a color format of the current video block is 4:4:4, and wherein look up tables and/or piece-wise linear models for chroma components are predictively coded in an adaptation parameter set (APS), a picture parameter set (PPS), a picture header, a slice header, a tile group header, a tile, or other video units.
65. The method of clause 64, wherein the look up tables and/or the piece-wise linear models of a color component are predicted from that of another color component.
66. The method of clause 58, wherein the method further includes: determining whether a color format of the current video block is 4:4:4, and wherein the look up tables and/or piece-wise linear models of a color component are derived from that of another color component.
67. The method of clause 66, wherein the look up tables and/or the piece-wise linear models of a chroma component are predicted from that of the luma component and/or another chroma component.
68. The method of clause 57, wherein the reshaping process is disabled and wherein reshaping for a chroma component of the current video block is directly disabled.
69. The method of clause 57, wherein the side information is not signaled.
70. The method of any of clauses 1 to 69, wherein the performing of the conversion includes generating the coded representation from the video.
71. The method of any of clauses 1 to 69, wherein the performing of the conversion includes generating the video from the coded representation.
72. An apparatus in a video system comprising a processor and a non-transitory memory with instructions thereon, wherein the instructions upon execution by the processor, cause the processor to implement the method in any one of clauses 1 to 71.
73. A computer program product stored on a non-transitory computer readable media, the computer program product including program code for carrying out the method in any one of clauses 1 to 71.
The third set of clauses describe certain features and aspects of the disclosed techniques listed in the previous section, including, for example, Example Implementation 3.
1. A video processing method, comprising: performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes side information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the side information provides parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the side information for a current video region is determined according to a rule that disallows inheritance or prediction of the side information based on a temporal layer index of the current video region.
2. The method of clause 1, wherein the inheritance or the prediction is disabled in a case that another picture is associated with a temporal layer index greater than the temporal layer index of the picture including the current video region.
3. The method of clause 1, wherein the inheritance or the prediction is only enabled in a case that another picture is stored in a decoded picture buffer.
4. The method of any of clauses 1 to 3, wherein one or more fields in reshaping information of another video region are copied to reshaping information of the current video region.
5. The method of clause 4, wherein the one or more fields include at least one of: i) reshaper_model_min_bin_idx, ii) reshaper_model_max_bin_idx, iii) RspCW[i], iv) RspDeltaCW [i], v) ReshapePivot[i], vi) ScaleCoef [i], vii) InvScaleCoeff [i], viii) ChromaScaleCoef [i], viiii) ix) tile_group_reshaper_enable_flag, or x) tile_group_reshaper_chr oma_residual_scale_flag.
6. A video processing method, comprising: performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes side information applicable for in-loop reshaping (ILR) of some of the one or more video regions, wherein the side information provides parameters for a reconstruction of a video unit of a video region based on a representation of the video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit, and wherein the side information for a current video region is determined according to a rule that disallows inheritance or prediction of the side information from a first type of another picture.
7. The method of clause 6, wherein the side information includes one or more lookup tables or parameters for one or more piece-wise linear models related to chroma samples, luma samples, or information enabling/disabling ILR, or information enabling/disabling chroma residual scaling.
8. The method of clause 6, wherein the current video region is a current picture, slice, tile group, or other video data units.
9. The method of clause 6, wherein the first type comprises pictures which are not reference pictures for the current video region including a current picture, slice, tile group, tile, or other video data unit.
10. The method of clause 6, wherein the inheritance or the prediction is enabled when a temporal layer index of the reference picture is no larger than a temporal index layer for the current video region including a current picture, slice, tile group, tile, or other video data unit.
11. The method of clause 6, wherein the first type comprises all pictures.
12. The method of clause 6, wherein the first type excludes a picture having smaller or equal temporal layer index of the current video region including a current picture, lice, tile group, tile, or other video data unit.
13. The method of clause 6, wherein the first type includes a picture having larger temporal layer index of the current video region including a current picture, lice, tile group, tile, or other video data unit.
14. The method of clause 13, wherein the picture is a reference picture for the current video region including a current picture, slice, tile group, tile or other video data unit.
15. The method of clause 6, wherein one or more reference lists and/or reference indices associated with another picture are signaled.
16. The method of clause 15, wherein the one or more reference lists and/or reference indices are included in a tile group header, a slice header, or a picture header.
17. The method of clause 16, wherein a syntax element includes a first field to indicate whether the side information is inherited from another reference picture in a given reference picture list and/or a second field to indicate a reference index of another reference picture from which the side information is inherited.
18. The method of clause 16, wherein the side information is initialized in a case that a reshaper model associated with the coding mode is not present in the tile group header and that the current video region is I-tile group or I-slice.
19. The method of clause 6, wherein the side information is inherited from the side information of a tile group or a slice of the reference picture.
20. The method of clause 6, wherein the reference picture is predefined.
21. The method of clauses 20, wherein the reference picture is a collocated picture.
22. The method of clause 20, wherein the reference picture has a reference index IDX and a reference list LX, and wherein LX=0 and IDX=0.
23. The method of clause 20, wherein the reference picture is closest in a displaying order to the current video region.
24. The method of clause 20, wherein the reference picture is closest in a decoding order to the current video region.
25. The method of any of clauses 1 to 24, wherein the performing of the conversion includes generating the coded representation from the video.
26. The method of any of clauses 1 to 24, wherein the performing of the conversion includes generating the video from the coded representation.
27. An apparatus in a video system comprising a processor and a non-transitory memory with instructions thereon, wherein the instructions upon execution by the processor, cause the processor to implement the method in any one of clauses 1 to 24.
28. A computer program product stored on a non-transitory computer readable media, the computer program product including program code for carrying out the method in any one of clauses 1 to 24.
The fourth set of clauses describe certain features and aspects of the disclosed techniques listed in the previous section, including, for example, Example Implementation 5.
1. A video processing method, comprising: performing a conversion between a coded representation of a video comprising one or more video regions and the video, wherein the coded representation includes a first side information at a first level, wherein a second side information at a second level is derived from the first side information such that the second side information provides parameters for a video unit coded with in-loop reshaping (ILR) in which a reconstruction of the video unit of a video region is based on a representation of a video unit in a first domain and a second domain and/or scaling chroma residue of a chroma video unit.
2. The method of clause 1, wherein the second side information is based on one or more syntax elements present in the second level.
3. The method of clause 1, wherein the first level is a tile group level and the second level is a tile level.
4. The method of clause 1, wherein the first level is a picture level and the second level is a tile group level.
5. The method of clause 1, wherein the first level is a picture level and the second level is a tile group level.
6. The method of clause 1, wherein the first level is an adaptation parameter set (APS) level and a second level is a picture level, tile group level, or tile level.
7. The method of clause 1, wherein the first side information is present in the ILR adaptation parameter set (APS) and the second side information is present in picture header.
8. The method of any clause of 1 to 7, wherein the second side information is inherited from the first side information.
9. The method of any clause of 1 to 7, wherein the second side information is predicted from the first side information.
10. The method of any clause of 1 to 7, wherein at least one of the first side information or the second side information comprises at least one of: information to enable or disable the ILR, a parameter of piece-wise linear models, information to enable or disable a reshaping between the first and second domain, or information to enable or disable a chroma residual scaling.
11. The method of any of clauses 1 to 10, wherein the performing of the conversion includes generating the coded representation from the video.
12. The method of any of clauses 1 to 10, wherein the performing of the conversion includes generating the video from the coded representation.
13. An apparatus in a video system comprising a processor and a non-transitory memory with instructions thereon, wherein the instructions upon execution by the processor, cause the processor to implement the method in any one of clauses 1 to 12.
14. A computer program product stored on a non-transitory computer readable media, the computer program product including program code for carrying out the method in any one of clauses 1 to 12.
In the present document, the term “video processing” or “visual media processing” may refer to image or video encoding, video decoding, video compression or video decompression. For example, video compression algorithms may be applied during conversion from pixel representation of a video to a corresponding bitstream representation or vice versa. The bitstream representation of a current video block may, for example, correspond to bits that are either co-located or spread in different places within the bitstream, as is defined by the syntax. For example, a macroblock may be encoded in terms of transformed and coded error residual values and also using bits in headers and other fields in the bitstream. Furthermore, during conversion, a decoder may parse a bitstream with the knowledge that some fields may be present, or absent, based on the determination, as is described in the above solutions. Similarly, an encoder may determine that certain syntax fields are or are not to be included and generate the coded representation accordingly by including or excluding the syntax fields from the coded representation. The side information may be, for example, an adaptation parameter set included with the coded representation or bitstream.
The disclosed and other solutions, examples, embodiments, modules and the functional operations described in this document can be implemented in digital electronic circuitry, or in computer software, firmware, or hardware, including the structures disclosed in this document and their structural equivalents, or in combinations of one or more of them. The disclosed and other embodiments can be implemented as one or more computer program products, i.e., one or more modules of computer program instructions encoded on a computer readable medium for execution by, or to control the operation of, data processing apparatus. The computer readable medium can be a machine-readable storage device, a machine-readable storage substrate, a memory device, a composition of matter effecting a machine-readable propagated signal, or a combination of one or more them. The term “data processing apparatus” encompasses all apparatus, devices, and machines for processing data, including by way of example a programmable processor, a computer, or multiple processors or computers. The apparatus can include, in addition to hardware, code that creates an execution environment for the computer program in question, e.g., code that constitutes processor firmware, a protocol stack, a database management system, an operating system, or a combination of one or more of them. A propagated signal is an artificially generated signal, e.g., a machine-generated electrical, optical, or electromagnetic signal, that is generated to encode information for transmission to suitable receiver apparatus.
A computer program (also known as a program, software, software application, script, or code) can be written in any form of programming language, including compiled or interpreted languages, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program does not necessarily correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, sub programs, or portions of code). A computer program can be deployed to be executed on one computer or on multiple computers that are located at one site or distributed across multiple sites and interconnected by a communication network.
The processes and logic flows described in this document can be performed by one or more programmable processors executing one or more computer programs to perform functions by operating on input data and generating output. The processes and logic flows can also be performed by, and apparatus can also be implemented as, special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application specific integrated circuit).
Processors suitable for the execution of a computer program include, by way of example, both general and special purpose microprocessors, and any one or more processors of any kind of digital computer. Generally, a processor will receive instructions and data from a read only memory or a random-access memory or both. The essential elements of a computer are a processor for performing instructions and one or more memory devices for storing instructions and data. Generally, a computer will also include, or be operatively coupled to receive data from or transfer data to, or both, one or more mass storage devices for storing data, e.g., magnetic, magneto optical disks, or optical disks. However, a computer need not have such devices. Computer readable media suitable for storing computer program instructions and data include all forms of non-volatile memory, media and memory devices, including by way of example semiconductor memory devices, e.g., EPROM, EEPROM, and flash memory devices; magnetic disks, e.g., internal hard disks or removable disks; magneto optical disks; and CD ROM and DVD-ROM disks. The processor and the memory can be supplemented by, or incorporated in, special purpose logic circuitry.
While this patent document contains many specifics, these should not be construed as limitations on the scope of any subject matter or of what may be claimed, but rather as descriptions of features that may be specific to particular embodiments of particular techniques. Certain features that are described in this patent document in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination. Moreover, although features may be described above as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed combination may be directed to a subcombination or variation of a subcombination.
Similarly, while operations are depicted in the drawings in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. Moreover, the separation of various system components in the embodiments described in this patent document should not be understood as requiring such separation in all embodiments.
Only a few implementations and examples are described and other implementations, enhancements and variations can be made based on what is described and illustrated in this patent document.
Zhang, Kai, Wang, Yue, Zhang, Li, Liu, Hongbin
Patent | Priority | Assignee | Title |
Executed on | Assignor | Assignee | Conveyance | Frame | Reel | Doc |
Feb 19 2020 | ZHANG, LI | BYTEDANCE INC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 057357 | /0810 | |
Feb 19 2020 | ZHANG, KAI | BYTEDANCE INC | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 057357 | /0810 | |
Feb 19 2020 | LIU, HONGBIN | BEIJING BYTEDANCE NETWORK TECHNOLOGY CO , LTD | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 057357 | /0860 | |
Feb 19 2020 | WANG, YUE | BEIJING BYTEDANCE NETWORK TECHNOLOGY CO , LTD | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 057357 | /0860 | |
Sep 01 2021 | BEIJING BYTEDANCE NETWORK TECHNOLOGY CO., LTD. | (assignment on the face of the patent) | / | |||
Sep 01 2021 | BYTEDANCE INC. | (assignment on the face of the patent) | / |
Date | Maintenance Fee Events |
Date | Maintenance Schedule |
Nov 12 2027 | 4 years fee payment window open |
May 12 2028 | 6 months grace period start (w surcharge) |
Nov 12 2028 | patent expiry (for year 4) |
Nov 12 2030 | 2 years to revive unintentionally abandoned end. (for year 4) |
Nov 12 2031 | 8 years fee payment window open |
May 12 2032 | 6 months grace period start (w surcharge) |
Nov 12 2032 | patent expiry (for year 8) |
Nov 12 2034 | 2 years to revive unintentionally abandoned end. (for year 8) |
Nov 12 2035 | 12 years fee payment window open |
May 12 2036 | 6 months grace period start (w surcharge) |
Nov 12 2036 | patent expiry (for year 12) |
Nov 12 2038 | 2 years to revive unintentionally abandoned end. (for year 12) |